Files
indd-utils/docs/format/fonts.md
T
aiongg eacc1be120 fix: name grouped styles with their group path and escape colons
A style's IDML Name is its group path and its own name joined by ":",
with a ":" inside a name written "\:" and "$ID/" before the whole path
of a built-in style; Self is that name escaped (objects.md, style
groups). Object and cell styles now include their group path too, and
every reference uses the same name. A font with an empty style name is
named "<family> Regular" with the family alone in Self (fonts.md).

compare.py, trustworthy pairs: extra values 33,181 -> 30,379 (0.27 %);
value coverage 92.45 % -> 92.50 % (10,307,358 values; grouped style
names, styles now matched); no value lost; 14 documents >= 99 %,
median 96.5 %. No conversion or schema failures; story text
18,509/18,610.
2026-10-07 22:34:08 +00:00

180 lines
9.0 KiB
Markdown
Raw Blame History

This file contains ambiguous Unicode characters
This file contains Unicode characters that might be confused with other characters. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.
# Fonts
Font families are objects of class 0x3E03. IDML writes them in
`Resources/Fonts.xml` as `FontFamily` elements with `Self="di<UID hex>"`,
each holding a `Font` element per font. Implemented in `src/model/font.rs`.
## Evidence
All corpus pairs, including those whose IDML is from an older version:
133 distinct INDD files with 630 font families that have an IDML
`FontFamily` with the same UID, holding 4,464 IDML `Font` elements. The
record layout below parses without error in all 1,354 font families of
the 251 distinct little-endian files.
## Family record (chunk 0x3E05)
| Field | Contents | IDML |
|---|---|---|
| u8, u16 | Not identified | |
| string | Family name | `FontFamily/@Name`, `Font/@FontFamily` |
| u8 | Not identified | |
| string | Family name in the font's own script | not written |
| 6 bytes | Not identified | |
| u16 | Font count | |
| font records | See below | one `Font` each, in the same order |
| u32 | Writing script | `WritingScript` of every font in the family |
Strings are in-object strings (`objects.md`).
**Font record:**
| Field | Contents | IDML `Font` attribute |
|---|---|---|
| u8 | Not identified (0 or 1) | |
| string | Style name | `FontStyleName` |
| u16 *n*, *n* bytes | PostScript name, one byte per character | `PostScriptName` |
| u8, string | Full name | `FullName` |
| u8, string | Style name in the font's own script | `FontStyleNameNative` |
| u8, string | Full name in the font's own script | `FullNameNative` |
| u32 | Font type: 1 `TrueType`, 6 `OpenTypeCFF`, 7 `OpenTypeCID`, 8 `OpenTypeTT` | `FontType` |
| u32 *n*, segments | Version: *n* UTF-16 code units as text segments | `Version` |
In files from InDesign 3.0 and 4.0 the PostScript name is a byte and an
in-object string, and InDesign 3.0 has no version (`big-endian.md`).
**Matches.**
- Font count: 629 of 630 families. The other family has 9 fonts in the
INDD and 4 in the IDML, which differ in their names, so the IDML was
probably exported from another save.
- Font order: the same in 627 of the 629 families. In the other two
(copies of one document), one font is at another position.
- In those 627 families, all 4,434 fonts match on `FontStyleName`,
`PostScriptName`, `FullName`, `FontStyleNameNative`, `FullNameNative`
and `Version`. In the other two, matched by name, 6 fonts of each have
another `Version` in IDML. That document's IDML gives its other font
families other UIDs, so it was probably exported from another save.
- `FontType`: 4,464 of 4,464 (code 6: 3,043, 1: 905, 7: 341, 8: 175).
Codes 0 (54 fonts) and 3 (1 font) occur in files without an IDML; the
converter leaves out `FontType` for them.
- A font with an empty style name: in all 21 such IDML fonts over all
pairs (12 in the trustworthy pairs), `Name` is the family name and
` Regular`, and `Self` is the family's `Self`, `Fontn` and the family
name, with no trailing space. All have `FontType="ATC"`.
- `WritingScript`: 4,464 of 4,464 with the u32 after the font records
(values 0, 1 and 33). Over all 654 pairs later in the corpus, 30,406
of 30,410 fonts; the other four, one family in one file, have 0 in the
INDD and 7 in the IDML.
One family in a file without an IDML stores `FF FF FF FF`. The schema
types `WritingScript` as `xsd:int`, and no corpus IDML has a negative
value, so the converter leaves the attribute out (with a warning) when
the u32 does not fit in an `int`.
- `Name` is the family name, a space and the style name, and `Self` is
the family's `Self`, `Fontn` and `Name` (4,464 of 4,464, taking the
family name as IDML writes it).
- `FontFamily/@Name`: 622 of 630. The other 8 have a technology suffix in
IDML that the INDD name lacks (`Montserrat (OTF)` 6 times, `Minion Pro
(OTF)` and `Trustpilot Display (OTF)` once each). A family whose INDD
name already ends in ` (TT)` keeps it. The converter writes the INDD
name, also in the fonts' `FontFamily`, `Name` and `Self`.
- IDML lists the families in UID order (135 of 135 pairs).
**Why the suffix is not predicted.** No field of the family record
separates the 8 from the other 622 families:
- The font type does not. The 8 hold OpenType CFF fonts (2) and
OpenType TrueType fonts (6), and both types occur in families without
a suffix: 130 other families with the name of the suffixed CFF family
hold the same CFF fonts, and 4 other `Montserrat` families the same
TrueType font.
- The first byte of the family record is 4 in 7 of the 8 and 1 in the
other, but 68 families without a suffix also have 4. One `Montserrat`
family with 4 has no suffix.
- A second family of the same name does not explain it. Only one of the
8 documents has one: two `Trustpilot Display` families, where the CFF
one has the suffix and the TrueType one has none. Other documents have
4 and 2 families of one name and type without a suffix.
- The other bytes of the record do not. The byte after the name is 0 in
the 8 and in 613 others. The first byte of each font record is 1 in
the 8 and in most others. The 6 bytes before the font count, read as
three u16, are (0, 0, 0xFFFF) or (0, 0xFFFF, 0xFFFF) in 7 of the 8,
as in 387 families without a suffix; the eighth has (0, 0, 2), and the
last u16 takes 12 different values in families without a suffix (2 in
4 of them), so it does not look like a flag.
The suffix most likely depends on the fonts installed where the IDML
was exported, as `Status` does (below).
Text formatting refers to a family by UID (`AppliedFont`, `BulletsFont`;
see `attributes.md`) and IDML writes the family name there.
## Typekit IDs (chunk 0x3EEB)
u32 count *n*, then *n* entries: u8 1 if the string is a built-in key
(written with `$ID/`), then a string. Entry *i* is the `TypekitID` of font
*i*; fonts after the *n*-th have `TypekitID="$ID/"`.
- 945 of 945 fonts in families where *n* equals the font count.
- In the 38 families where *n* is smaller, the entries match the first
*n* fonts and all other fonts have `$ID/` (38 of 38).
- Families with *n* = 0 or without the chunk: all 3,020 fonts in IDML
from DOM 12 on have `$ID/`. The one DOM 7 IDML (32 fonts) has no
`TypekitID`; the converter writes it for InDesign 12 and later.
## Not converted
- `Status` (`Installed`, `Substituted`, `NotAvailable`) depends on the
fonts present where the IDML was exported. No INDD field agrees with
it: the first byte of the font record is 1 for 2,042 `Installed` and
1,047 `Substituted` fonts. The attribute is optional in the schema and
left out.
- `NumDesignAxes`, `DesignAxesName` and `DesignAxesValues` (131 fonts of
variable font families) are not in the records above, and were not
found elsewhere.
## Composite fonts
Every INDD file has one object of class 0xCB02, and every IDML one
`CompositeFont`, `[No composite font]` (240 of 240 IDML files). Its
entries are objects of class 0xCB03; IDML writes them as
`CompositeFontEntry` with `Self="u<UID>"`, six per file. Implemented in
`src/model/cjk.rs`.
**Composite font (chunk 0xCB02).** A flag byte (1 = built-in key, `$ID/`)
and the name, fields not identified, then a u16 count and the UIDs of
the entries, which end the chunk. The list is the IDML entry order in
76 of 78 pairs; the other two are pairs whose IDML was exported from
another save (their entry UIDs have no INDD object). No sample has
another composite font, so the converter writes only this one.
**Entry (chunk 0xCB03).**
| Field | IDML | Evidence |
|---|---|---|
| Flag byte, string | `Name` | 456 of 456 |
| u32 font family UID | `AppliedFont` (Properties): the family's name | 454 of 456; the other 2 are in the pair whose IDML names the family `Minion Pro (OTF)` (above) |
| Flag byte, string | `FontStyle` | 456 of 456 (`$ID/R`, `$ID/Regular`, `$ID/Roman`) |
| Four f64 | `RelativeSize`, `HorizontalScale`, `VerticalScale` 100, `BaselineShift` 0 | (100, 0, 100, 100) in all 1,500 entries of the 250 distinct little-endian files |
| u16 1, u16 count *n*, *n* ranges | `CustomCharacters` | 380 of 380 |
| Four u16 | `ScaleOption`: all 1 `true`, all 0 `false` | 456 of 456 |
Each range is three code points (first, last, first again), each one
UTF-16 unit or a surrogate pair. `CustomCharacters` is every character
from first to last of each range, in order. IDML never writes
`CustomCharacters` for the `$ID/Kanji` entry (240 of 240 files), whose
ranges cover most of Unicode; the converter leaves it out for that entry.
In one file without IDML the `$ID/kCompFontString_Base` entry covers all
of Unicode (U+0000–U+D7FF and U+E000–U+10FFFF); XML 1.0 does not allow
U+FFFE and U+FFFF, so the converter leaves those two characters out of
every attribute value and text.
The four numbers have one value in every sample, so only the 0 is told
apart (`BaselineShift`); the converter writes the four attributes only
when the numbers are (100, 0, 100, 100). Which of the four u16 is
`ScaleOption` is not known; the converter writes it only when all four
agree. `Locked="true"` is in every IDML entry and is written from that
observation (`idml-values.md`).
`PlatformName` is `$ID/` in all 4,464 IDML fonts; the converter writes
that value (`idml-values.md`).